AI Use Policy

I use AI tools to build faster, and anyone hiring me should be able to read the rules before they think to ask for them. So here they are, dated and public. What you are hiring is judgment I have spent years building. The tools handle the mechanical half, I decide every part of the work, and my name is on the result either way.

The rules below are the guardrails. This is why they are worth having.

It is why my prices look the way they do

BC agencies commonly quote fifteen to fifty thousand dollars for builds I deliver for a fraction of that. Their price carries a team and an office. Mine carries one person with sharp tools, and the difference lands in your budget.

It buys care that used to be uneconomic

Every page checked at every screen size, accessibility and speed measured rather than assumed, every link tested, every price swept for consistency across the whole site. Being that thorough on a small project used to cost more than the project. Now it is routine, and it is the difference between a site that works and one that mostly works.

Nothing of yours gets quietly trimmed

When migration is expensive, corners get cut and clients are told a selection of their content will move over. Your whole archive comes across, laid out to be read.

The saved hours go back into your project

They go into your design, your words, and understanding how your business actually works. No tool can do that half, and it is the half you are paying me for.

Rule 1: You know where AI sits before you hire me, and personal information stays out of it

AI is part of my toolchain in the same way hosting, a code editor, and design software are, and every agreement says so in plain words before it is signed. Nobody discovers it afterward. What I hold back is personal information: data about identifiable people, your customers, your donors, your mailing list. That does not go into an AI tool on a general permission, because it is yours to control and the law treats it differently. If a job ever calls for it, I come to you about that specific thing and you can say no.

Rule 2: No deliverable ships on AI's word

Every deliverable is verified by me against the project's quality gates before it reaches you: code is reviewed and type-checked, rendered output is checked on real screen sizes, copy is edited by hand, accessibility and performance are measured. AI output that has not been verified is a draft, and drafts do not ship.

Rule 3: Accountability never delegates

I am the builder of record on every project. AI compresses the mechanical share of the work; it does not replace judgment, design decisions, or ownership of the result. When you call, the person who answers is the person responsible, and "the tool did it" will never be a sentence you hear from me.

Rule 4: The honesty rules apply to the machines too

My standing rules ban invented statistics, fabricated reviews or testimonials, synthetic clients, and fake case studies. Concept work is labelled as concept work everywhere it appears. AI makes fabrication cheap, which is exactly why the ban applies regardless of what produced it.

Rule 5: Ask me anything

Any client can ask where AI was used in their project and get a plain answer. Generated imagery I use is art-directed and post-processed by me, and I will tell you which images those are if you want to know.

It handles the mechanical half of a build: bulk code changes, verification sweeps, research, and first drafts I then edit and often rewrite. The design direction, the judgment calls, and the words that carry my name stay mine, and those are what a client is actually buying.

Processing your customers' personal information without a specific approval for that specific job. Producing "proof" of anything. Replacing the verification steps that make the work trustworthy. Auto-publishing anything client-facing without review. These are standing rules, and the workflow is built around them.

Separate from my own workflow, I also build AI features into client sites where they serve visitors. Those ship with the same governance: guardrails, cost controls, and honest limits stated up front. If a feature cannot be made trustworthy, it does not ship.

I also build sites to be AI-answer ready: structured, schema-rich, and machine-readable, so that when someone asks an AI assistant about your business, your own site is the clearest source available. The honest limit: nobody can promise how a third-party assistant behaves, and I will never pretend otherwise.